Can you explain how this managed to be a 15,600 instead of a frame slower? I edited out both the level transition frames, and Virtualdub says it's 937 frames / 60 FPS (that's what I record in) = 15,61666666666667

When the player first spawns (the frame you first see the level) is when the time starts. This is considered frame 0. This is how the Speedrun in-game timer works too and how the ILs have been timed. The frame before the level transitions (the last frame you see the level) is when the timer ends.

Your video starts on the pre-level message screen and it's there for only a frame (or maybe 2 if you're watching in 60FPS, I was only able to download the 30FPS version). The frame after that, frame 1, is when the timer begins. You then complete the level by overlapping the end green area 15 seconds and 19 frames into the video, the final frame of the video. This means the time for the run is 15s 19frames (end time) - 1frame (start time) giving us 15 seconds and 18 frames. We convert frames into milliseconds and we've got the 15.6s you see on the leaderboards.

Recordings above 30FPS are not factored into the timing and only help ensure no frames are missed by recording at a higher framerate than what the game runs.

You also had some concerns about hsblue's level 11 run. This one is a bit tricky. The video starts a frame late, which can be seen by the position of the player on the first frame in the video. This is why your timing found the run to be 1 frame faster than what it actually is.

There's also another issue with the video: the last frame of the level is missed. You can see the frame right before he touches the end goal, but not the frame where he is overlapping. I believe this is due to a desync between the recording and the game, as my copy of the video appears to be 29.97 FPS (although YouTube tells me it's running at 30). You can also see two blue dots on-top of the 'U' in 'HURRY' as evidence of recording artifacts. There's also a lot of duplicated frames and frame jumps in the run. You can see when it's 11.24 on the in-game timer it's there for 2 frames before going to 11.25. Might have made the run a frame slower on recording. Also goes from 11.00 on one frame to 11.02 on the next and 11.03 the one after, potentially 'catching up' to the game it's recording. Not sure about this one, just went off the in-game timer.

Right, so your method of timing matches mine; I start on frame 0 and edit out the beginning transition screen, and stop and on the frame before the second transition screen. That comes out as 937 frames on the non-transcoded 60 FPS video, and if I download the 60 FPS video off youtube (I use JDownloader 2) and similarly edit out the transition screens, the result would also be 937 frames/9,617s

This is the 60 FPS video: https://mega.nz/#!7YlD2I4B!TXClKXEa01Gw8rA3mKOeH6gK9FhOuOsf7Z337wwGpiU Perhaps timing the 30 FPS version results in a discrepancy?

Thanks for the raw video. I checked it out and got the same results as before. Starts with the transition for 2 frames on the recording (frames 0 and 1) and then the level starts on the third frame (frame 2). Player overlaps ending 15 seconds and 38 frames into the video (frame 938). Total time is (938 - 2) / 60 = 15.6s

I think what's happened here to get 937 frames total is the 60FPS causing some confusion. Recording the game at 60FPS will capture every game frame twice. You've marked the end as the frame right before the end transition appears (frame 939), whereas I've used the frame the player first overlapped the finish (frame 938). These are the same frame in the game, but have been recorded twice due to 60FPS.

I think you're right, Pzula - however, there's another oddity that's worth checking out: I highly recommend comparing our positioning in both the videos. I think you will find that our positions are exactly the same when compared against the cycle at the end of the stage. If I'm a frame faster (a 30 FPS frame), then there'd be a noticeable difference, but our paces are identical. I believe the in-game timer lied and your time is actually 15,6s